David Newman
Well-known member
Andreas said:You put me in to serious thoughts here. The way you put it it makes a LOT of sense that we should be resizing before we start editing. I really mean that it sounds like a MAST DO. What are your suggestion, comments for someone like me that is won by Canopus and planning to edit native DVCPRO-HD ? We somehow been over this before, but please explain a little more, are you suggesting that using connect-HD from the beggining and converting-resizing our footage will give better quality in the long run, plus the overall boost up ?
You realy put me in some serious thoughts here... Please analyze the most that you can...
A lot of this agrument relates the native versus intermediate approach to post production. Native is good it you don't intend to enhance the data, no color correction, no titles, just cuts, and master out to the source footage (in this operation it lossless, even if the source is lossly.) All non-cut operations are lossy in native, and more so than intermediate solutions as the source is not optimized for post (that is a long discussion in itself.) There really very few pro-video productions that benefit from native source format editing. It is sometimes argued that as long as you avoid renders (no compositing it, or use of external tools) the native is OK when master to another format (but there are many ifs.)
-- VS --
In HD uncompressed is often used as an intermediate, it is ideal in quality, not so ideal in data rate, but it is a reference point because everyone agrees on the quality. In an uncompressed/intermediate workflow, source files are compressed (which compression back to source is lossly -- native champions may even point this makes native better than uncompressed, silly really) yet uncompressed has no issue with generation loss in post. Now with a 960x720 source, no one will work uncompressed at 960x720, they will upres upon decode to 1280x720 (some even to 1920x1080.) They are very good reasons for this, 1) square pixels, 2) HDSDI only operates in standard HD resolutions -- i.e. if you don't scale in post, your HD-SDI card (or Deck in the case of DVCPRO-HD decks) will scale the image for you, 3) all you post elements have more resolution. Upscaling to 1280x720 gives you 33% more pixel data, more luma and chroma resolution. And the important 1:1 pixel match to all your output formats, HDSDI, D5, broadcast, HD DVD and BluRay. This is not agrument that CineForm is a way to go, this is true for any NLE.
Now that I've the argument where an intermediate is better, it is a smaller step to realize a compressed intermediate is better (and faster and cheaper because the smaller file sizes.) Of course if the new compression quality is bad, stick with native (or uncompressed), but if the new compression is better than the source compression, and allows for 1:1 pixel formats in 4:2:2, there is no downside.
Now to push to issue further. An 8-bit source can benefit for a 10-bit workflow. For any intermediate renders (like After Effects work) 8-bit processing can introduce banding, so 10-bit compression is the solution (or 10-bit uncompressed.) CineForm offers a compressed 10-bit real-time workflow (Prospect HD) and the only other doing this is AVID, so we are good company. If you read up on AVID's DNxHD you will see they are saying the same things. They even show the advantages DNxHD has over DVCPRO-HD just as CineForm does.
I can go on and on.
Last edited: